Machine learning models predict MOOC certificate purchasing with accuracies of 0.82–0.91 using only time spent on each course step¶

Claim · All claims
Evidence · 1 study · associational r2 · q2
Currency · newest study 2021 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Four ML algorithms (RandomForest, GradientBoosting, AdaBoost, XGBoost) predicted course purchasability from step-level time-spent features with accuracies between 0.82 and 0.91. → Alshehri 2021

Evidence¶

Alshehri 2021¶

Alshehri, M., Alamri, A., Cristea, A. I., & Stewart, C. D. (2021). Towards Designing Profitable Courses: Predicting Student Purchasing Behaviour in MOOCs. International Journal of Artificial Intelligence in Education, 31, 215–233. https://doi.org/10.1007/s40593-021-00246-2

q2 · i? · associational · r2

Predictive modelling study on FutureLearn clickstream data from 23 runs of 5 University of Warwick courses, evaluating four algorithms with train/test split and 10-fold cross validation. The authors report "promising accuracies, between 0.82 and 0.91" using time spent per step alone; no effect size is printed.

"Our proposed model achieved promising accuracies, between 0.82 and 0.91, using the time spent on each step only."

Discussion¶

Related Claims¶

• Boosting models (XGBoost) achieved the highest predictive accuracy among the compared ML models, while logistic regression showed the lowest — related
• Baseline LR and XGBoost models achieve comparable predictive performance on OULAD, with XGBoost slightly higher AUC-ROC — reports the opposite
• MOOC certificate purchasing is rare: under 1% of enrolled learners purchase, with completion around 10% — related
• The run number (delivery iteration) is among the most representative features for predicting MOOC purchasing behaviour — related